Volumes

The Volumes view lists Docker volumes and — on Business Edition — lets you manage them across the whole swarm. The read-only view is part of the base TUI; listing volumes on all nodes and the create / delete / prune / file-browse actions are license-gated and require a managed context.

Open it from the command bar:

:volume

What you can do depends on two things:

• License. The volumes-all-nodes entitlement (granted on the be and trial tiers — see License) unlocks cross-node listing and all management actions.
• Context. Cross-node operations talk to the per-node agents through the RBAC proxy, so they need a managed Docker context (a swarm that has been put through :bootstrap). On a non-managed context the view falls back to the connected node only.

The Volumes view

The list has six columns:

Column	Meaning
NAME	Volume name.
STACK	Owning stack (from the com.docker.stack.namespace label), or -.
DRIVER	Volume driver (local, …).
MOUNT POINT	On-disk path of the volume on its node.
CREATED	Creation timestamp.
HOST	The node the volume lives on.

Navigation and inspection:

Key	Action
↑/↓ (k/j)	Move cursor
pgup / pgdown	Page up / down
i / enter	Inspect the selected volume (JSON viewer)
/	Filter (matches name, stack, driver, mount point, host)
?	Full help
esc / q	Back

Sort with shift+ the column initial; press the same key again to flip the direction:

Key	Sort by
shift+n	Name
shift+s	Stack
shift+d	Driver
shift+c	Created
shift+h	Host

Scope: connected node vs. all nodes

Without the license, the view lists volumes on the connected node only and shows a footer hint:

Connected node only · listing volumes on all nodes is a Business Edition feature: …

With volumes-all-nodes licensed on a managed context, the view aggregates volumes from every node in the swarm (the HOST column tells you where each one lives). Two degraded states surface as a non-fatal banner rather than an error:

• Not bootstrapped (licensed, but the context has no managed proxy): Connected node only · cross-node volume management needs :bootstrap.
• Some nodes unreachable: the reachable volumes are still listed, with an "N nodes unreachable" banner — the listing degrades, it does not fail.

Managing volumes

The management actions are license-gated. On an unlicensed or non-managed context the keys still exist but open a "Business Edition feature" dialog instead of acting.

Key	Action
c	Create a volume
b	Browse files in the selected volume
ctrl+d	Delete the selected volume
p	Prune unused volumes

Create (c)

Opens a form with a Name field and a Node picker (the swarm's nodes). tab cycles focus, the arrow keys choose the node, enter creates, esc cancels. Volumes are created with the local driver.

Delete (ctrl+d)

Asks for confirmation — Delete volume '<name>' on <host>? This cannot be undone. — then removes the volume on its node. y/enter confirms, n/esc cancels. Deleting a volume that is still mounted by a container fails with an "in use" error; stop the consumer first.

Prune (p)

Removes unused volumes (docker volume prune --all). Pick a single node, or toggle Prune all nodes with space to run it across the whole swarm. After confirming, the result dialog reports how many volumes were deleted, how much space was reclaimed, and any per-node errors.

Browse files (b)

Opens an in-volume file browser rooted at the volume's contents:

Key	Action
↑/↓ (k/j)	Move cursor
enter / →	Enter directory / select file
p / backspace / ←	Up to parent directory
r	Rename the selected entry
ctrl+d / delete	Delete the selected entry
d	Download to your local machine
u	Upload from your local machine
esc	Back

Download (d) streams the selected entry to a local path: a file is written raw, a directory as a .tar. The transfer streams with live progress, refuses to overwrite an existing destination, and writes to a temporary file that is atomically renamed on success.

Upload (u) is the inverse — it opens a local file browser (a [Upload this directory] row selects the directory you are browsing). A file is sent raw into the current directory; a directory is packed and unpacked under it. Uploads are capped by the agent's FILEOP_MAX_UPLOAD_BYTES (default 2 GiB); exceeding it returns 413. Operators change the ceiling in the bootstrap stack and redeploy — see Configuration.

File operations run inside a short-lived, network-isolated helper container scoped to the target volume; they do not run on the agent host's filesystem.

How cross-node access works

Volume operations travel over the same authenticated, authorized path as the other Business Edition features: swarmcli asks the RBAC proxy, the proxy checks your role, and the request reaches the node holding the volume.

Listing is aggregated across every node, so you see the whole swarm's volumes in one view; every other operation is addressed to the node you selected. A node that is unreachable is reported as such rather than silently omitted.

Permissions and gating

volumes-all-nodes is granted on the be and trial tiers (see License — Model). When it is off — no license, an expired or wrong-swarm license, or a non-managed context — the view shows the read-only connected-node listing and the management keys open the "Business Edition feature" dialog. Role-based access to the underlying calls is enforced centrally by the RBAC proxy; see RBAC.

Failure modes

What you see	Cause
"Business Edition feature" dialog on c/b/ctrl+d/p	No volumes-all-nodes entitlement, or a non-managed context. Open :license, or switch to a bootstrapped context.
Connected node only · cross-node volume management needs :bootstrap	Licensed, but the active context has no managed proxy. Run :bootstrap.
"N nodes unreachable" banner	Some agents did not answer in time; the listing shows the reachable nodes. Check :bootstrap --check.
Delete fails with an "in use" error	The volume is still mounted by a running container. Remove or stop the consumer, then retry.
upload exceeds the maximum size of <N> bytes (413)	The upload is larger than FILEOP_MAX_UPLOAD_BYTES. Raise the limit in the bootstrap stack and redeploy, or split the upload.